The Biological Underpinnings of Namib Desert Fairy Circles

See allHide authors and affiliations

Science  29 Mar 2013:
Vol. 339, Issue 6127, pp. 1618-1621
DOI: 10.1126/science.1222999

Fairies? No, Termites!

Fairy circles consist of circles of perennial vegetation that grow within otherwise mostly barren desert habitat on the southwest coast of Africa. Many hypotheses have been put forward to explain the creation and maintenance of fairy circles. Using long-term data collected on the distribution and physical and biological components of these features, Juergens (p. 1618) found that the circles are generated by the actions of the sand termite, which removes vegetation produced following intermittent rains. Once generated, the circles collect water, which sustains the growth of perennial vegetation at the edges of the circles, allowing for long-term persistence of the termites.


The sand termite Psammotermes allocerus generates local ecosystems, so-called fairy circles, through removal of short-lived vegetation that appears after rain, leaving circular barren patches. Because of rapid percolation and lack of evapotranspiration, water is retained within the circles. This process results in the formation of rings of perennial vegetation that facilitate termite survival and locally increase biodiversity. This termite-generated ecosystem persists through prolonged droughts lasting many decades.

Fairy circles (FCs) are large, conspicuous, circular patches devoid of vegetation in the center but with perennial grasses at the margin. These patches occur in large numbers in the desert margin grasslands of southern Africa (Fig. 1, A and B). Early observers considered poisonous plants, ants, or termites as causal factors; however, most of these early hypotheses were systematically tested and rejected (1, 2). It has also been proposed that an unknown semivolatile substance in the soil might be responsible for the absence of grass within the FCs (2, 3). In fact, a wide range of volatile organic compounds are found in FCs (4). Measurements of carbon monoxide and hydrocarbons in the soil led to the proposal of a geochemical origin of FCs (5). Carnivorous ants (6) and “self-organizing vegetation dynamics” (7) have also been considered as causes for FCs. Despite the many hypotheses, the origin and the ecosystem function of FCs are still a much-debated mystery. I used a long-term data set describing the environmental and biogeographical characteristics and dynamics of FCs to identify the most likely cause of these unique formations. Additionally, I analyze the function of FCs in terms of water management, biodiversity, and adaptation to arid conditions.

Fig. 1

(A) Spatial pattern of FCs. (B) Geographical distribution of FCs (black dots) and hotspots of FC occurrences at wider landscape scale (yellow clusters). Note the proximity of the occurrences to the 100-mm isohyet. (C) Schematic sketch showing the structural elements within a FC. The PB is formed by a ring of much larger perennial plants located around the bare patch. The grass outside the FC dies and disappears during drought. The PB remains the only surviving vegetation biomass during drought periods.

FCs occur along a narrow belt at the eastern margin of the Namib Desert, running from mid-Angola to northwestern South Africa. The area of distribution is closely associated with the isohyet of 100-mm mean annual precipitation (MAP) (Fig. 1B). The disjunct occurrence of FCs is caused by their pronounced restriction to sandy soils.

High soil humidity within FCs has been observed previously (1, 2). To confirm and quantify this potentially adaptive function, I measured volumetric soil water content (m3/m3 × 100) from 2006 to 2012 within and around FCs. At sites with a MAP of 100 mm, more than 53 mm of water were stored in the upper 100 cm of soil, even during the driest time of the year (table S1). At a depth >40 cm, a soil humidity of more than 5% volumetric water content was recorded over 4 years.

Higher temporal resolution of water flux was gained by automatic measurements recorded every hour within the bare patch and the grass matrix at 10-, 30-, 60-, and 90-cm depths using FDR sensors. During the observation period of 4 years, the humidity at 60-cm depth within the FC was either at or well above 5% volumetric water content (Fig. 2A). In the typical sand texture of FC soils with dominant grain sizes around 180 μm and pore sizes around 50 μm, 5% volumetric water content causes more than 98% relative air humidity in the pore system.

Fig. 2

(A) Rainfall events and volumetric soil water content (volume percent, m3/m3 × 100) at different depths underneath a FC, measured hourly from early 2008 until mid-2010. (B) Same measurements as in (A) comparing the bare patch (FC, solid lines) and the matrix (MT, thin lines) measured from 15 May 2011 until 6 October 2012.

I hypothesize that the FC, while not losing water by transpiration because of the absence of plants in its center, accumulates rain water as a result of the rapidly draining large pore size of sand. The rapid percolation to a deeper soil layer reduces evaporation loss. Simultaneous measurements taken under the matrix vegetation showed much lower water contents (Fig. 2B). The amount and the longevity of the water body underneath the FC allow the formation of a belt of perennial grasses at its margin. Their roots extend only 20 to 30 cm into the bare patch. The majority of FCs possess such a perennial belt (PB), and they are essential for the ecosystem functioning of the FCs. Here, I test the hypothesis of a biogenic origin by interpreting the area of distribution of FCs as a clearly defined environmental envelope (80 to 120 mm MAP, deep sandy soils) of an organismic taxon causing intraspecific competition for space (2).

Data collected over 40 field trips were used to systematically assess which organisms are associated with FCs. At each site (Fig. 1B), at least 30, and in some cases more than 100, FCs were investigated above and below ground.

Species distribution maps show that only a few organisms are associated with FC hotspots across their entire distribution (table S2). Among termites, only the sand termite (Psammotermes allocerus) was found at all FC hotspots, whereas Hodotermes mossambicus is largely restricted to the summer rainfall climate and Microhodotermes viator is limited to the winter rainfall climate. Baucaliotermes hainsii only occurs south of the southern Central Namib. P. allocerus is widely distributed over southern Africa and thus exceeds the FC distribution. Three ant species—Messor denticornis, Anoplolepis steingroeveri, and Tetramorium sp.—were found in several FC hotspots, but none of them in all.

If direct presence of organisms in or next to all single FCs (table S2) is scrutinized, only P. allocerus was found in high frequencies (80 to 100%). The characteristic “sheetings” (thin layers of cemented sand built over the foraged plant material) of P. allocerus (fig. S2, A, C, and D) were found at 80 to 100% of the FCs and throughout all life stages of the FCs. In addition, in 80 to 100% of FCs, P. allocerus nests (Fig. 3B and fig. S12) and underground tunnel-like galleries with a characteristic black organic wall covering (tapetum) (figs. S9C and S12, C and E) were found a few centimeters to decimeters underneath the bare patch, the PB, and the matrix area. The frequency of these observations was halved during the wet season.

Fig. 3

(A) Activity pattern of P. allocerus in a 1-m2 grid within and outside a FC at Giribesvlakte, expressed as density of soil dumps per m2. The maximum figures within the bare patch, a minimum at the PB, and a secondary maximum in the matrix outside the PB indicate the foraging activities in the surrounding area. Counts in the matrix are random samples, and empty squares have not been counted. (B) Example of spatial distribution of foraging nests (round pictograms) and permanent surface nests (rectangular pictograms) (nests with living termites colored red; abandoned nests, ocher). The green ring marks the PB. (C) Assessment of 83 FCs with regard to the number of living grass plants (individuals) found in the bare patch, plotted against the termite activity, and measured as average number of soil dumps per m2 within the bare patch.

Although these associations suggest a causal role for P. allocerus, it is possible that they may instead merely reflect the colonization of FCs by the termites. However, sand termites were found even in the initial state of new FCs, that is, before the water accumulation has begun and the perennial grass belt has developed. Careful assessment of 24 newly formed FCs at Giribesvlakte in Namibia in March 2012 revealed the presence of P. allocerus in all of them. In these youngest FCs, the dying grass plants were damaged only at the roots, associated with underground galleries of P. allocerus (fig. S9C). No other organism has been observed foraging on the grass of young FCs.

During the further life history of FCs, P. allocerus is directly involved in keeping the bare patch of FCs free of grass. The related presence and activity of P. allocerus can be best assessed at night and in the morning, when the workers clean the underground burrows and create characteristic small soil dumps (Fig. 3A and fig. S10). Within a random stratified sample of 83 FCs at Giribesvlakte, a negative correlation was found between the density of the soil dumps at the bare patch and the number of surviving grass plants (Fig. 3C). These correlations suggest that the burrowing activities of P. allocerus within the bare patch do not only serve in taking up water (8); their foraging on the roots of freshly germinated grasses kills them and keeps the bare patch free of vegetation. Furthermore, P. allocerus is involved in widening the diameter of the circle. During most of the FCs’ adulthood, the termites steadily feed on a few (often neighboring) perennial grass plants at the inner margin of the PB (figs. S2A, S11D, and S12B), thereby slowly widening the diameter of the FC. Out of a total of 160 FCs examined at Giribesvlakte in March 2011, 96% showed remains of dead grass tussocks at the margin of the bare patch. At 53% of these FCs, some of the grass tussocks—on average, 24.9 (minimum of 1, maximum 80)—were still covered with P. allocerus sheetings.

The main ecosystem function of FCs is related to securing two important perennial long-term resources. First, the removal of all water-transpiring plants allows the accumulation of water underneath the FC after rain events (water trap). I hypothesize that the generation of a perennial water supply facilitates the survival of termites in a hostile desert. Whereas the annual rainfall evenly distributed in space allows ephemeral or annual plant growth, the removal of plants allows perennial growth of plants in the PB. I argue that this generation of perennial plant biomass is the second facilitator of survival of termites, even in extreme drought years. The manner in which the termites create and manage the perennial grass population within an otherwise ephemeral desert environment supports the hypothesis of active ecosystem “engineering.” The formation of the PB is a consequence of the water accumulation and the unidirecional suppression of competition, both caused by the termites.

FCs strongly enhance biodiversity by attracting many organisms. Evidence of this was established by comparing lists of taxa observed in and near FCs with lists established in nearby grasslands without FCs. A number of ants, bees, wasps, small mammals, and plants are found more often in and near FCs. Often the mainly granivorous ant M. denticornis establishes itself in the center of the bare patch and forages along linear foraging trails in the wider surroundings of the FC. Plant species, for example, the Cucurbitaceae Citrullus lanatus with its large water-storing fruits and even Acacia erioloba trees, establish themselves within or next to FCs in the reticulate dunes at the eastern margin of the Namib dune field. Furthermore, the population of P. allocerus termites itself forms an attractive resource, which is used by geckos (Palmatogecko rangei), aardvarks (Orycteropus afer), bat-eared foxes (Otocyon megalotis), black-backed jackals (Canis mesomelas), golden moles (Eremitalpa granti), and spiders (e.g., Seothyra) as well as by omnivorous ants like Anoplolepis steingroeveri (6) and Tetramorium sp., which have been regularly observed attacking P. allocerus workers. In summary, FCs, like oases in the desert, increase biodiversity (quantified as the number of species) by one to two orders of magnitude (table S3).

FCs can be regarded as an outstanding example of allogenic ecosystem engineering resulting in unique landscapes with increased biodiversity, driven by key resources such as permanently available water, perennial plant biomass, and perennial termite biomass. The termites match the beaver (9) with regard to intensity of environmental change, but they surpass it with regard to the spatial dimension of their impact. P. allocerus turns wide desert regions of predominantly ephemeral life into landscapes dominated by species-rich perennial grassland (Fig. 1A and figs. S1C, S6, and S7), supporting uninterrupted perennial life even during dry seasons and drought years.

Supplementary Materials

Supplementary Text

Figs. S1 to S12

Tables S1 to S3


References and Notes

  1. Acknowledgments: Funding of long-term monitoring was provided by the German Federal Ministry of Education and Research 01 LC 0024, 01 LC 0024A, and 01 LC 0624A2 [Biodiversity Monitoring Transect Analysis in Africa (BIOTA)–Southern Africa]. Data presented in this paper are available within the supplementary materials.
View Abstract

Navigate This Article